Carrier Ethernet. Chapter. by Abdul Kasim

Transcription

1 MH Technical title / Delivering Carrier Ethernet: Extending Ethernet Beyond the LAN / Kasim / / Chapter 2 Chapter 2 Carrier Ethernet by Abdul Kasim In order to leverage the potential of Ethernet beyond the LAN, it had to be augmented with additional carrier-class characteristics; identifying and formalizing these detailed characteristics was, therefore, essential to enabling this role for Ethernet. This chapter focuses specifically on standardization and other efforts underway to develop a foundation for transforming LAN Ethernet into a Service Provider based offering, henceforth referred to as Carrier Ethernet (services). Carrier Ethernet delivered over Service Provider networks across the MAN and WAN optimally enables next-generation packet applications. The first fundamental step is defining Carrier Ethernet, what it precisely means and understanding the rationale for this definition. Also as fundamental, is an established reference framework the context in which this definition applies, and the necessary elements that make up this context. In so doing, a common and consistent understanding as well as a language to describe Carrier Ethernet services is provided; with this as the basis, the attributes are discussed in greater detail (note: in the context of this book, only a sufficient overview can be reasonably provided), with selective discussions in a few areas that are deemed especially critical to enabling Carrier Ethernet. Most of the standardization effort, especially at the service-level, has been carried on by the Metro Ethernet Forum (MEF) and so expectedly, this chapter devotes a significant part to the MEF-initiated development; but efforts by other standards bodies are also identified. This chapter also attempts to incorporate some commercial developments enabling Carrier Ethernet. Often, forward-looking entities whether Service Providers or equipment manufacturers are ahead of the standards bodies in terms of recognizing and addressing the practical issues that usually emerge when offering new services. A look at these issues and their respective solutions in the marketplace serves, therefore, to provide a better understanding of the actual status quo in the field. 45 ch02.indd 45 10/1/07 10:31:48 PM

2 46 Chapter 2 Defining Carrier Ethernet Although numerous efforts, both informal and formal (standards-based), have been undertaken to make Ethernet more viable as a technology and service beyond the LAN, the MEF has been instrumental in initiating a substantial formal effort to define Carrier Ethernet services (delivered by Service Providers). This definition was a prerequisite to developing a common understanding and a common objective in the delivery of such services. Among the first steps undertaken was to define more precisely what such Ethernet services would entail, since, as noted in the previous chapter and repeated in Table 2.1, there are fundamental differences in providing Ethernet in the Service Provider network (broadly referred to as Carrier Ethernet) as opposed to providing Ethernet in the LAN. The context in which Carrier Ethernet services are defined is, therefore, the Service Provider networks and the several types of services already being delivered over these TABLE 2.1 Ethernet in the LAN Versus Ethernet in a Service Provide Network (Spanning the MAN and WAN) Dimension Local Area Network Service Provider Network Geography/Reach Service Provider Usually less than 1 2 km; deployed in building(s) and small campuses Enterprise (IT group); implemented by internal IT group. User of service Enterprise Enterprise km and longer; deployed in a metro area or even across distant metro areas Service Provider (Carrier typically); services offered commercially for an initial and recurring cost Number of end users/points (Scale) In the tens/hundreds Thousands or tens/hundreds of thousands Bandwidth 10M/100M/1000M 1M and greater up to 10,000M; usually in granular increments of 1M Aggregation required Services offered (scope) Enterprise data applications Voice/ TDM and data connectivity applications such as Internet Access, intra-metro connectivity Delivery of Ethernet services Tolerance to failures (resiliency) Manageability Over coax (CAT 5) and fiber; Best effort Generally reasonable because network is usually intra-enterprise and over a smaller physical area so failures can be addressed relatively quickly Manageability possible with fairly simple tools given fewer number of users and applications within a smaller physical area (typically a building or campus) and the relatively higher tolerance to failure issues Over a host of media, incumbent transport technologies, and with an associated service-level agreement (SLA) Very low tolerance because failures usually have a larger impact often on revenues and competitiveness Scale and scope of the Service Provider network in terms of the number of users and the geographical footprint introduces significant complexity necessitating sophisticated management tools and capabilities ch02.indd 46 10/1/07 10:31:48 PM

3 Carrier Ethernet 47 networks. In fact, Carrier Ethernet essentially encompasses the deterministic and other service delivery aspects for standardized Ethernet services. This point is key because it highlights the focus on standardized Ethernet services and the specific characteristics of such services and not necessarily the underlying transport infrastructure itself. So what is Carrier Ethernet? Carrier Ethernet: A Formal Definition The MEF 1 has defined Carrier Ethernet as the ubiquitous, standardized, Carrier-class service defined by five attributes that distinguish Carrier Ethernet from the familiar LAN based Ethernet. As depicted in Figure 2.1, these five attributes, in no particular order, are 1. Standardized services 2. Scalability 3. Reliability 4. Quality of Service (QoS) 5. Service management Carrier Ethernet essentially augments traditional Ethernet, optimized for LAN deployment, with Carrier-class capabilities which make it optimal for deployment in Service Provider Access/Metro Area Networks and beyond, to the Wide Area Network. And conversely, from an end-user (enterprise) standpoint, Carrier Ethernet is a service that not only provides a standard Ethernet (or for that matter, a standardized non-ethernet 2 ) handoff but also provides the robustness, deterministic performance, management, and flexibility expected of Carrier-class services. Fundamental to both Carrier Ethernet and LAN Ethernet is the fact that data is carried in an Ethernet frame. What this means is, in effect, an Ethernet frame originating at a device in the LAN, now continues to traverse across one or more Service Provider networks, 3 largely unaltered, and terminates at a device in a remote LAN. One way to look at this transformation is that it essentially creates one larger Ethernet, spanning LANs, MANs, and may be even the WAN, albeit delivered as a service to the customer. This transformation is shown in Figure 2.2, courtesy of the MEF, and illustrates the remarkable potential of Carrier Ethernet. The terms and NNI in the figure denote standardized interface hand-offs between the enterprise customer and 1 MEF is the preeminent nonprofit industry body focused solely on enabling Carrier Ethernet. The Metro reference in MEF is now a misnomer, however, and does not accurately reflect its charter and focus, which has long extended beyond the metro. 2 Because it can, as will be seen later, also support non-ethernet services (albeit over an Ethernet layer). 3 The Service Provider networks could encompass both the MAN and the WAN. ch02.indd 47 10/1/07 10:31:49 PM

4 48 Chapter 2 Standardized Services Quality of Service Carrier Ethernet Scalability Service Management Reliability Figure 2.1 Attributes of Carrier Ethernet (Source: MEF) Video Global/National Carrier Ethernet Metro Carrier Ethernet Access Carrier Ethernet Ethernet Services Eth Layer Internet Subscriber Site Service Provider 1 Metro Ethernet Network E-NNI Service Provider 2 Metro Ethernet Network Subscriber Site Subscriber Site Subscriber Site Figure 2.2 Carrier Ethernet spanning Access, Metro, and Wide Area Networks (Source: MEF) ch02.indd 48 10/1/07 10:31:50 PM

5 Carrier Ethernet 49 the Service Provider network and between Service Providers (or Network Operators 4 ), whose infrastructure is used to deliver the service, respectively, and are explained in more detail later in this chapter. The Ethernet frame(s) may be transported as is, either natively and directly over a physical media or encapsulated and delivered over a variety of overlay networks built using different technologies. Each of these very different networking technology solutions, however, delivers 5 Carrier Ethernet services. It is critical to understand that the Carrier Ethernet attributes often manifest only partially in commercial solutions today because they exist at the network/transport/physical layers as opposed to the service layer 6. This will become clear in rest of the Part II when the various commercial solutions currently employed to deliver Carrier Ethernet are discussed. NOTE The focus in this book is primarily on delivering Carrier Ethernet services; the network and transport delivery infrastructure the Carrier Ethernet solutions, provide the carrier-class attributes that enable commercial Carrier Ethernet services. Often, the term Carrier Ethernet is interchangeably used to refer to both the Ethernet services and the underpinning enabling solution infrastructure. The Carrier-class attributes are delivered differently by the various network solutions (for example, how reliability is offered in one solution versus another). This is largely a result of their respective geneses and subsequent evolution. It is important to also note that some of the Carrier Ethernet attributes in a solution existed pre-carrier Ethernet (albeit at the transport layer and not at the service layer) and were, in fact, initial drivers for the use of respective solution. For example, SONET offered impressive resiliency to any failures in the fiber and/or equipment deployed in a ring topology, so it was adopted to support mission-critical voice services that required stringent SLAs. Each of the Carrier Ethernet solutions and its respective evolution toward optimizing delivery in Service Provider networks is discussed in a fair amount of detail in Part II of the book. Carrier Ethernet: The Attributes The five attributes that define Carrier Ethernet essentially provide the additional capabilities necessary to use Ethernet in much the same way as the other preceding service provider technologies such as ATM and Frame Relay. 7 Each of these attributes is elaborated upon and its rationale highlighted in the sections that follow. 4 A Network Operator is distinguished from a Service Provider by the fact that the former s infrastructure is employed in the delivery of Carrier Ethernet services; however, the service itself is commercially offered to the customers (usually on a subscription basis) by the Service Provider. Service Provider often lease infrastructure from network operators to deliver services. 5 More accurately, as will be evident in Part II, the solutions strive to offer the attributes of Carrier Ethernet. 6 Because at these lower layers inherently address only a subset of the higher-level service. 7 Especially helpful today because Ethernet is largely being used as a substitute for Frame Relay and ATM. ch02.indd 49 10/1/07 10:31:50 PM

6 50 Chapter 2 Standardized Services This attribute essentially enables a Service Provider to deliver a host of both Packet and traditional TDM (see chapter 10 for more information on TDM) multi-point services in an efficient and deterministic manner over standardized equipment platforms. These services underpin the multitude of customer applications that are emerging across voice, data, and video. Specific components that define this attribute comprehensively are defined next. Ubiquity Carrier Ethernet enables ubiquitous Ethernet services provided via standardized equipment, independent of the underlying media and transport infrastructure. This is a critical prerequisite to extending Ethernet s appeal globally (similar to LAN Ethernet). Ethernet Services Carrier Ethernet supports two types of services: Point-to- Point (also referred to as Ethernet Line or E-LINE) and multipoint-to-multipoint Ethernet LAN (referred to as E-LAN) Ethernet services. These services are discussed in greater detail later in the chapter and are expected to provide the basis for all Ethernet services. Circuit Emulation Services (CES) Carrier Ethernet supports not only Ethernet-based services delivered across different transport technologies but also other (TDM) services transported over Carrier Ethernet itself. As noted previously, TDM services still remain an overwhelming contributor to Service Provider revenues and realistically need to be supported (and delivered over a converged Ethernet-based infrastructure). TDM-based voice applications especially need to be accommodated and characteristics of such applications such as synchronization and signaling need to be emulated. Granularity and Quality of Services (QoS) The services supported by Carrier Ethernet provide a wide choice and granularity of bandwidth and quality of service options. This flexibility is vital in Service Provider networks with its multitude of end users, each with slightly different application requirements and, typically, operating equipment from multiple vendors. QoS capability is crucial to enforcing the deterministic behavior of Carrier Ethernet. Converged transport Supports convergence of voice, data, and video services over a unified (Ethernet) transport and greatly simplifies the delivery, management, and addition of such services. Basically, all enterprise services and applications are now supported over a single Ethernet pipe. Scalability One fundamental difference between a LAN and a Service Provider network 8 is scale. In a Service Provider network, there are usually a hundredfold more end users and as a consequence, exponentially more connections for Ethernet-based 8 Or multiple Service Provider or Network Operator networks, since several such entities could be involved in the delivery of an Ethernet service. A Network Operator owns the delivery infrastructure but may or may not be the one offering a service (or the Service Provider). ch02.indd 50 10/1/07 10:31:50 PM

7 Carrier Ethernet 51 applications simply because it covers a larger geographic area. Carrier Ethernet solutions, therefore, scale across several dimensions simultaneously: Users/endpoints A Service Provider network supports hundreds of thousands of endpoints and millions of Ethernet users in an optimal fashion. Specifically, it supports the delivery of millions of Ethernet services with an appropriate level of performance or QoS. Geographical reach The services delivered can span access, metros, and beyond to encompass very large geographical distances and over a variety of infrastructures including Ethernet, WiFi, WiMax, TDM, SONET, and so on. As noted previously, the reach of such services can be augmented by employing multiple Service Providers adjacent networks. Applications Current and emerging applications supporting a host of business, information, and entertainment applications and benefiting from the convergence of voice, data, and video. The landscape or breadth of application support is a vital driver for Carrier Ethernet. Bandwidth Bandwidth scales from 1M to 10G in granular increments of 1M, enabling a much more palatable solution to both the end user and Service Provider because end users only have to pay for what is required and Service Providers would possibly receive higher revenues. As these dimensions scale collectively, they make for a formidable problem to deliver, isolate, troubleshoot, and in general, manage thousands of users and hundreds of thousands of services in a robust manner. Reliability As Carrier Ethernet services are expected to support mission-critical applications on a wide scale, the ability to detect quickly and remotely any failures that may arise in the physical infrastructure or in the Ethernet services layer underlying these applications is essential. Specifically, the following aspects are addressed by Carrier Ethernet. Service Resiliency The impact of failures is localized and will not affect other customers and/or applications; Correlation among multiple errors will be quickly identified. Further, the process of troubleshooting and recovery from failures will be rapid and employ tools that will minimize operational expenditures for the Service Provider and any adverse impact on the end users. Protection Carrier Ethernet services provide an end-to-end service-level protection that encompasses protection against any failures in the underlying infrastructure employed in the delivery of the services. This means protection against failures in the end-to-end path of the service, as well as against any underlying physical link and node equipment failures. Restoration Carrier Ethernet provides similar or better recovery than SONET. The benchmark for resiliency in Service Provider networks has long been the ch02.indd 51 10/1/07 10:31:51 PM

8 52 Chapter 2 SONET sub-50 ms service restoration to support circuit-switched voice networks. As latency-sensitive voice and video applications are deployed over a Carrier Ethernet infrastructure, this SONET-like resiliency is a critical prerequisite. Techniques such as Spanning Tree Protocol (STP) and its variants, while feasible in the LAN, are simply not acceptable in large Service Provider networks because depending on the size and complexity of the network, recovery of failures employing these techniques takes in the range of several seconds to even minutes. Carrier Ethernet supports a host of latency-sensitive applications that are often critical to an enterprise (for instance, regular telephony services), and consequently offers better fault-tolerant and recovery mechanisms. Quality of Service Providing Quality of Service (QoS) is necessary for Carrier Ethernet to be embraced as a substitute to ATM and Frame Relay and ultimately as a converged mechanism to deliver all services. QoS essentially conforms to a predefined level of performance expected by an application. As Carrier Ethernet supports delivery of critical enterprise applications that are commonly expected to adhere to certain performance levels, this QoS capability becomes essential. The challenge to a Service Provider is significant given the fact that it has to simultaneously support individual QoS to typically thousands of applications and end users, using a limited set of resources (bandwidth, switching, and so on) whose availability varies with time. Carrier Ethernet services providing QoS, encompass the following: Performance Service Level Agreement (SLA) There is the capability to provide the stringent end-to-end 9 SLAs necessary to provide a host of critical voice, video, and data services over a converged Ethernet infrastructure. Such SLAs are essential, and end users often demand them since they are already accustomed to such an assurance using the ATM, Frame Relay, or Private Line services, and it is only natural for them to expect the same of Ethernet services that support similar and next-generation applications. SLA parameters A set of configurable parameters allows a Service Provider to actually define the specific SLAs associated with a particular commercial service. These parameters provide significant latitude for defining numerous levels of service premiums. Further, these parameters although associated with a service, are enforced across the underlying infrastructure delivering that service. Provisioning SLA The QoS provides a hard performance guarantee based on the typical elements that define QoS in networks such as availability at a particular performance, packet loss, packet delay, and packet delay variation or jitter. In a LAN with its abundant bandwidth and high performance, QoS is usually not an issue; the simple priority queuing capability using IEEE 802.1P/802.1D provides a soft 9 End-to-end refers to the end points between which an Ethernet service is delivered. ch02.indd 52 10/1/07 10:31:51 PM

9 Carrier Ethernet 53 QOS, but this is not sufficient in a Service Provider network, where with a multitude of users competing for shared resources (bandwidth, switching, and so on), complexity is at a totally different level. Different techniques, therefore, become necessary. Service Management Managing a large number of customers stretched over a wider geographical area requires Service Providers to have a sophisticated capability for installing, troubleshooting, and upgrading Ethernet services cost effectively and quickly; engaging in a truck-roll each time there is an issue is simply cost-prohibitive and makes it infeasible to deliver Ethernet on a wider scale. Carrier Ethernet, in an attempt to address these issues, provides. Unified management This encompasses standardized vendor-independent capability to monitor, diagnose, and manage the delivery infrastructure. It is not unusual to deliver services across multiple Service Provider networks, each of which is often comprised of equipment from one or more manufacturers and is frequently subject to individual differences; hence, managing services across the different vendors equipment using a common streamlined approach becomes paramount. Carrier-class OAM Carrier-class Operational, Administration, and Maintenance (OAM) capability that will integrate with existing Service Provider operational models. This covers a wide array of capabilities that enables life-cycle management at the service level. With Carrier Ethernet based networks reaching tens of kilometers and thousands of subscribers, the need for sophisticated OAM features is apparent. Carrier Ethernet incorporates cutting-edge service creation and management techniques that exceed those of both enterprise Ethernet and the legacy telecom infrastructure. Rapid Provisioning The capability to provision new Ethernet services rapidly is a key departure from the long and protracted commissioning intervals for traditional TDM services. This capability translates into allowing granular increases in bandwidth to existing services; the addition of new services, each with a specific performance assurance (SLA); and the ability to enable these services remotely most of the time. Carrier Ethernet leverages the established benefits of LAN Ethernet to the end users while simultaneously enabling Service Providers to offer a set of carrier-class attributes in a manner that is not only aligned with other services such as ATM, Frame Relay, and Private Line, but does so in a scalable, robust, and flexible manner that supports the next-generation of packet-based applications much more cost effectively. This ultimately translates into lower CAPital EXpenditures (CAPEX), lower OPerational EXpenditures (OPEX), and competitive positioning for Service Providers. Thus, Carrier Ethernet helps realize the compelling benefits to both end users and Service Providers as detailed in the Chapter 1. Defining the attributes of Carrier Ethernet in greater detail and refining them further to be more relevant for next-generation applications is an ongoing effort; considerable progress has, however, been made. ch02.indd 53 10/1/07 10:31:51 PM

10 54 Chapter 2 Enabling Carrier Ethernet Carrier Ethernet is increasingly being adopted by the Service Provider and enterprise end user community not only as the default access solution (i.e., service connectivity is via Carrier Ethernet), but also one that is being employed end-to-end across the WAN. Service Provider networks are, in fact, evolving to deliver the consistent Carrier-class Ethernet services end users are coming to expect. Chapter 3 highlights the growing demand for Carrier Ethernet services worldwide. Carrier Ethernet is, however, still a work in progress; in fact, it is still in its infancy and being more formally defined, refined, and continually augmented based on learning from real life field deployments supporting emerging applications. If it is to achieve the success and dominance of its LAN variant, it has to not only incorporate these lessons rapidly in terms of new value-added features, but also standardize them. Standardization of Carrier Ethernet is thus a key approach to enabling and, in fact, accelerating the deployment of Carrier Ethernet services. Standards Bodies There are several standard bodies that are involved, to varying degrees, in enabling Carrier Ethernet. These include the IEEE (primarily the 802 body), the Internet Engineering Task Force (IETF), the International Telecommunications Union (ITU), the Metro Ethernet Forum (MEF), and to a lesser degree, others such as the Tele Management Forum (TMF). While the involvement of several bodies working in the same area may appear to be at cross purposes or at best, partially redundant and with the potential to introduce confusion, the reality has been different. These bodies have been and are working with a largely complementary focus, and where there has been some overlap, there has also been significant collaboration, with the net result actually expediting standardization efforts. The IETF has traditionally had an IT orientation, while the ITU has focused on developing international standards to support the needs of national Service Providers (known as PTTs in most countries). The IEEE, of course, has focused on the 802 Ethernet standards at the physical and data-link layer. It is continuing its legacy work on Ethernet and extending it in two areas from the standpoint of Ethernet in the MAN and WAN: OAM and Architecture. The ITU is working across the spectrum, from service definition to service architecture to OAM and Ethernet interfaces. These bodies were involved with LAN Ethernet and are now also focused on Carrier Ethernet given its role as a converged platform appealing to both Service Providers and end-user enterprises and spanning their traditional Service Provider and IT constituencies. The Metro Ethernet Forum (MEF), unlike the others, was formed relatively recently (2001) and exclusively to advance the deployment of Carrier Ethernet. Consequently, it has been the most active body focused on enabling Carrier Ethernet as a well-defined service to support the next-generation of applications. And although the MEF s initial focus was the delivery of Carrier Ethernet in the metropolitan area ch02.indd 54 10/1/07 10:31:52 PM

11 Carrier Ethernet 55 (hence the Metro in MEF), it has now extended its charter well beyond and focuses on end-to-end Carrier Ethernet services spanning the MAN and the WAN. The MEF represents the first comprehensive effort to address all service delivery aspects as well as the testing necessary for confirmation. Figure 2.3 depicts the different MEF standards and their respective focus as of August 2007; these are continually being augmented as the MEF tackles new issues in its attempt to accelerate the deployment of Carrier Ethernet. While the MEF has a broader mandate than the other bodies (at least as far as Carrier Ethernet is concerned), it extensively builds and reuses the efforts of these bodies. Figure 2.4 summarizes the different standards bodies respective focus across four distinct areas with respect to Carrier Ethernet: Architecture, Services, Management, and Testing. The specific standards are identified in each of these areas; standards underway but not yet ratified are italicized. It is clear that only the MEF is focused across the board in all four areas and is notably the only standards body testing and validating Carrier Ethernet. The IEEE 802 addresses some architectural aspects (in fact, it did so even pre-carrier Ethernet) but has also added several new efforts. A key contribution has been in the area of Carrier Ethernet Management, especially link level and connectivity management. The ITU has been very active in the Architecture, Service, and Management areas; it has not only leveraged the efforts from the other standards bodies for instance, the MEF for the Ethernet services definition, the IEEE for Management but has also augmented it by, for instance, adding Performance Management to its Management standard to address the requirements of its constituency. Specification MEF 2 MEF 3 MEF 4 MEF 6 MEF 7 MEF 8 MEF 9 MEF 10.1 MEF 11 MEF 12 MEF 13 MEF 14 MEF 15 MEF 16 MEF 17 MEF 18 MEF 19 Scope Requirements and Framework for Ethernet Service Protection Circuit Emulation Service Definitions, Framework and Requirements in Metro Ethernet Networks Metro Ethernet Network Architecture Framework Part 1: Generic framework Metro Ethernet Services Definitions Phase 1 EMS-NMS Information Model Implementation Agreement for the Emulation of PDH Circuits over Metro Ethernet Networks Abstract Test Suite for Ethernet Services at the Ethernet Services Attributes Phase 2 User Network Interface () Requirements and Framework Metro Ethernet Network Architecture Framework Part 2: Ethernet Services Layer User Network Interface () Type 1 Implementation Agreement Abstract Test Suite for Ethernet Services at the Requirements for Management of Metro Ethernet Phase 1 Network Elements Ethernet Local Management Interface Service OAM Framework and Requirement Abstract Test Suit for Circuit Emulation Services Abstract Test Suit for Type 1 Figure 2.3 MEF Standards specifications (Source: MEF) ch02.indd 55 10/1/07 10:31:53 PM

12 56 Chapter 2 Ethernet Standards Summary Standards Body IEEE Ethernet Services - Architecture/Control MAC 802.3ar Congestion Management 802.1D/Q Bridges/VLAN RPR 802.1ad Provider Bridges.1ah Provider Backbone Bridges (PBB).1ak Multiple Registration Protocol.1aj Two Port MAC Relay.1AE/af MAC / Key Security.1aq Shortest Path Bridging.1Qay PBB Traffic Engineering Ethernet OAM 802.3ah EFM OAM 802.1ag CFM 802.1AB - Discovery 802.1ap VLAN MIB Ethernet Interfaces PHYs 802.3as - Frame Expansion MEF MEF 10 Service Attributes MEF 3 Circuit Emulation MEF 6 Service Definition MEF 8 PDH Emulation MEF 9 Test Suites MEF 14 Test Suites Services Phase 2 MEF 4 Generic Architecture MEF 2 Protection Req & Framework MEF 11 Req & Framework MEF 12 - Layer Architecture MEF 7 EMS-NMS Info Model MEF 15 NE Management Req OAM Req & Framework OAM Protocol Phase 1 Performance Monitoring MEF 13 - Type 1 MEF 16 ELMI E-NNI ITU G.8011 Services Framework G EPL Service G EVPL Service G.asm Service Mgmt Arch G.smc Service Mgmt Chnl G.8010 Layer Architecture G.8021 Equipment Model G.8010v2 Layer Architecture G.8021v2 Equipment Model Y.17ethmpls - ETH-MPLS Interwork Y.1730 Ethernet OAM Req Y.1731 OAM Mechanisms G.8031 Protection Y.17ethqos QoS Y.ethperf - Performance G.8012 / NNI G.8012v2 / NNI TMF - - TMF814 EMS to NMS Model - Figure 2.4 Standards bodies and their respective areas of Carrier Ethernet focus (Source: MEF) The detailed specifications are, of course, vastly outside the scope of this book but they are referenced in sufficient detail in the context of defining Carrier Ethernet services and its underlying five attributes. This represents the formalized (i.e., standardization) effort thus far toward enabling Carrier Ethernet. It must be noted that Carrier Ethernet standardization activity is relatively dynamic and frequently there continues to be new developments. It is, therefore, advisable to check the websites of the standards bodies (see bibliography) to get a sense of the latest progress. A Service Architecture for Carrier Ethernet Since Carrier Ethernet is essentially a commercial service offered by a Service Provider, it was vital to establish a clear and precise specification of what it entails. This was especially necessary because Ethernet in the LAN was not typically offered as a service but rather as a product/solution wherein the equipment was purchased, set up, and managed by the enterprise (IT group) itself. As such, there were generally no serviceoriented expectations of the LAN Ethernet. Unfortunately, this was also largely the case with the Ethernet services that were initially (and to some extent are still being) offered by Service Providers. There were no formalized definitions and expectations of these services. The effort to change this only recently began in earnest and has been driven primarily by the MEF. Although still in the beginning stages, reasonably significant progress has been made. Even before formalizing Carrier Ethernet services, however, it was necessary to establish a context for such services a Service architecture and to identify the necessary service components of such an architectural context. The MEF (and also the ITU) ch02.indd 56 10/1/07 10:31:54 PM

13 Carrier Ethernet 57 undertook this effort and developed a set of standard specifications for a generic Service Architecture that provides a common language for describing Ethernet services. In MEF 6 and MEF 10.1, the MEF has established what an Ethernet service is, how a variety of subscriber services can be offered, and how these Ethernet services can be customized for certain performance and Service-Level Agreements (SLAs). The MEF has also defined an overall framework to discuss Ethernet services the Ethernet Service Model (ESM), which identifies the building blocks or service attributes of these services. (The Ethernet Service Model does not define the Ethernet service itself; this is done in an Ethernet Service Definition framework explained later.) The Ethernet Service Model (ESM) The basic Service Provider architectural model defined by the MEF is shown in Figure 2.5. It has two main components: The Subscriber or customer equipment (CE) The Metro Ethernet Network (MEN) or more accurately, the Service Provider Ethernet Network (SEN) 10 This is owned/operated by a Service Provider. Basically, the customer equipment is connected to the MEN [through a User Network Interface () which is explained in greater detail in the next section]. Any OSI layer 1 or 2 transport technology can be used as long as Ethernet frames are being handed off. The Subscriber or customer equipment is typically a router or a switch (an IEEE 802.1Q bridge). A MEN itself consists of physical components (e.g., network elements, ports, etc.) and logical components (e.g., meters, policers, shapers, virtual switches, links, etc.). It can be owned and operated by multiple Service Providers and provides the underlying transport (SONET, WDM, RPR, etc.) to carry the Ethernet frames. It essentially connects geographically separated enterprise LANs across the MAN and WAN. The Carrier Ethernet service is actually provided by the Service Provider owning the MEN over an Ethernet Virtual Connection (or EVC, which is defined in a later section). The MEF has more formally defined a three-layered model (also shown in Figure 2.5) for the MEN; the Application services (APP) layer supports end-user applications carried over Ethernet connectivity services provided at the Ethernet services (ETH) layer, and these connectivity services in turn are delivered over various transport/networking technologies in the Transport services (TRAN) layer. The key focus of the MEF and other standards bodies is the ETH layer; Carrier Ethernet is defined in this layer. The delivery of these Carrier Ethernet services can be over various media and the transport and networking technologies that make up the TRAN layer (the subject of Part II). 10 Since the MEF has extended its focus beyond the metro and into the WAN, it is generally more accurate to label the Metro Ethernet Network (MEN) as the Service Provider Ethernet Network (SEN), which could support the MAN and/or WAN. ch02.indd 57 10/1/07 10:31:54 PM

14 58 Chapter 2 Application Services Layer (e.g., IP, MPLS, PDH, etc.) Ethernet Services Layer (Ethernet Service PDU) Data Plane Control Plane Management Plane Transport Services Layer (e.g., IEEE 802.1, SONET/SDH, MPLS) Subscriber Site A Subscriber Site B Client T Network Metro Ethernet Network (MEN) Network T Client Ethernet Virtual Connection (EVC) End-to-End Ethernet Flow Figure 2.5 The basic Service Provider model for delivering Ethernet services (Source: MEF) As will become evident in Part II, often the current and evolving attributes of Carrier Ethernet reside in the TRAN layer (depending on the specific technologies). Each of the three layers has three associated operational planes: a Data plane, a Control plane, and a Management plane. The Data plane, also referred to as the user/transport/forwarding plane, provides the functional elements required to steer the subscriber flow and supports the transport of subscriber traffic units among MEN Network Elements (NEs). The Control plane provides the functional elements that support distributed flow-management functions among NECs participating in the MEN data plane. The Control plane also provides the signaling mechanisms necessary to support distributed setup, supervision, and connection release operations, among other flow-control functions. The Management plane provides the functional elements that support Fault, Configuration (including flow and/ or connection configuration), Account, Performance, and Security (FCAPS) functions, as well as any related Operations, Administration, and Maintenance (OAM) tools. The three operational planes are generally well defined for the TRAN layer (numerous standards bodies have addressed it, and these are identified in Part II). For the ETH layer, the effort was, for the most part (except in the data plane), begun only recently. As will become evident in the rest of the book, the control and management functions of the TRAN layer are often employed in delivering Carrier Ethernet currently. ch02.indd 58 10/1/07 10:31:55 PM

15 Carrier Ethernet 59 Ethernet services delivered over the MEN invariably have two key service attributes associated with them: the User Network Interface () and the Ethernet Virtual Connection (EVC). User-Network Interface () The is the interface used to interconnect a subscriber to an Ethernet Service Provider. The also provides a reference point for demarcation between the MEN operator s (i.e., a Service Provider s) equipment that enables access to the MEN services and the subscriber access equipment. The demarcation point indicates the location where the responsibility of the Service Provider ends and where the responsibility of the subscriber begins. The is a key Ethernet service attribute used to specify an Ethernet service. Functionally, the is an asymmetric, compound functional element that consists of a client side, referred to as the -C, and a network side, referred to as the -N. Thus, the term is used to refer to these two functional elements and generically, to the data, management, and control plane functions associated with them. Client (-C) The -C represents all of the functions required to connect a subscriber to a MEN. Individual functions in a -C are entirely in the subscriber domain, and may or may not be managed by the Service Provider/Network Operator. From the perspective of the MEN, the -C supports the set of functions required to exchange data, control, and management plane information with the MEN subscriber. As such, the -C includes functions associated with the Ethernet services infrastructure, the transport network infrastructure, and if present, application-specific components. Network (-N) The -N represents all of the functions required to connect a MEN to a MEN subscriber. The individual functions in a -N are entirely in the Service Provider/Network Operator domain. From the perspective of the subscriber, the -N supports the set of functions required to exchange data, control, and management plane information with the MEN. As such, the -N includes functions associated with the Ethernet services infrastructure, the transport network infrastructure, and if present, application-specific components. The MEF has defined a set of attributes to specify a completely. These are listed at the end of the chapter (Figure 2.24). Ethernet Virtual Connection (EVC) The Ethernet Virtual Connection (EVC) is a construct that performs two functions: One, it indicates the association of two or more s for the purpose of delivering an Ethernet flow 11 between subscriber sites across the MEN. Two, an EVC prevents data transfers between subscriber sites that are not part of the 11 An Ethernet flow represents a particular and potentially noncontiguous (e.g., consecutive Ethernet frames may belong to different flows) unidirectional stream of Ethernet frames that share a common treatment for the purpose of transfer steering across the MEN. ch02.indd 59 10/1/07 10:31:55 PM

16 60 Chapter 2 same EVC. The attributes associated with an EVC are shown in Figure 2.24 (at the end of the chapter) and are employed when specifying an Ethernet service. NOTE There may be one or more subscriber flows mapped to a particular EVC. This capability enables an EVC to provide data privacy and security. There are two basic rules that govern the delivery of Ethernet frames over an EVC. A service frame must never be delivered back to the where it originated, and the Ethernet frame contents (including MAC addresses) must remain unchanged. The MEF has defined two types of EVCs: Point-to-Point or Multipoint-to-Multipoint. In a Point-to- Point EVC, exactly two s must be associated with one another whereas in a Multipointto-Multipoint EVC, two or more s must be associated with one another. Thus, an EVC can be used to construct a Layer 2 Private Line or a Layer 2 VPN 12 service. Network to Network Interfaces (NNI) As noted in the reference Service Architecture, one or more Service Providers can be used to deliver Carrier Ethernet services. The demarcation or handoff between the Service Providers is referred to as the Network-to- Network Interfaces (NNIs). The MEF has defined several NNIs: External Network-to-Network Interface (E-NNI) An open interface used to interconnect two MEN Service Providers. Internal Network-to-Network Interface (I-NNI) An open interface used to interconnect network elements from a given MEN Service Provider. Network Interworking Network-to-Network Interface (NI-NNI) An open interface that supports the extension of transport facilities used to support Ethernet services and associated EVCs over an external transport network not directly involved in the end-to-end Ethernet service. Service Interworking Network-to-Network Interface (SI-NNI) An interface that supports the interworking of an MEF service with services provided via other service enabling technologies (e.g., Frame Relay, ATM, IP, etc.). Defining Carrier Ethernet Services Carrier Ethernet services are essentially connectivity services that employ Ethernet frames transported over the MEN using a host of different technologies such as SONET, WDM, MPLS, and so on. As shown in Figure 2.5, Ethernet services are delivered over an EVC provided by a Service Provider over a MEN, which is connected to the customer equipment (CE) via a standardized. Thus, all Ethernet services will invariably have associated with them, one or more s and one or more EVCs. The specific and EVC attributes differentiate the specific services. 12 Virtual Private Network (VPN) is a connectivity service between multiple points to multiple points. ch02.indd 60 10/1/07 10:31:55 PM

17 Carrier Ethernet 61 Defining Carrier Ethernet Services MEF Ethernet Service Definition Framework Carrier Ethernet Service Defined by Ethernet Service Type Defined by Ethernet Service Attribute Defined by Ethernet Service Attribute Parameters & EVC Attributes Associated with the Service Type Carrier Ethernet Services E-Line Ethernet Physical Interface Traffic Parameters Performance Parameters Class of Service Actual Values for Each of the /EVC Attributes E-LAN Service Frame Delivery E-Tree* VLAN Tag Support Service Multiplexing Bundling * Not ratified yet Security Filters Figure 2.6 Defining Ethernet services Carrier Ethernet services are defined from a subscriber perspective (and hence they re also referred to as retail services ). As shown in Figure 2.6, the MEF has developed an Ethernet Services Definition Framework that defines any Carrier Ethernet service in terms of a predefined Ethernet service type. Each of these Ethernet service types (described next) are, in turn, defined by a set of Ethernet service attributes that define its capabilities. Some of these attributes apply to the, others to the EVCs, and still others to both the and EVCs associated with the service type. Specific parameters associated with each of these Ethernet service attributes ultimately define the Ethernet service fully. This seemingly complicated approach is also illustrated in Figure 2.6, but it will become clearer when real-life examples are discussed later in the chapter. It is helpful to remember that every service is defined in terms of a service type and invariably has a set of and EVC attributes 13 that will uniquely define it. Before delving into the specific service types (which are defined in terms of the Ethernet service attributes), it is useful to understand these service attributes. 13 Collectively referred to as the set of Ethernet service attributes. ch02.indd 61 10/1/07 10:31:56 PM

18 62 Chapter 2 Ethernet Service Attributes The Ethernet service attributes are categorized into the following groups: Ethernet Physical interface, Traffic parameters, Performance parameters, Class of Service, Service frame delivery, VLAN tag support, Service Multiplexing, Bundling, and Security filters. Whether they apply to only the or EVC or both is identified in the brief descriptions that follow. Ethernet Physical Interface At the, the Ethernet physical interface has several service attributes. Physical Medium This service attribute specifies the physical interface defined by the IEEE standard. Examples are 10BaseT, 100BaseSX, 1000BaseLX, and so on. Speed This service attribute specifies the standard Ethernet speed either 10 Mbps, 100 Mbps, 1 Gbps, or 10 Gbps. Mode This service attribute specifies whether the supports full or half duplex 14 and can provide auto-negotiation. MAC Layer This service attribute specifies which MAC layer is supported, i.e., as specified in the IEEE Traffic Parameters/Bandwidth Profile The MEF has defined the Bandwidth Profile service attribute, which is associated with every Ethernet service and can be applied at the or for an EVC. When there are multiple services associated with a, there is a corresponding Bandwidth profile associated with each of these services. A Bandwidth profile specifies a limit on the rate at which Ethernet frames can traverse the associated with an Ethernet service. Bandwidth profiles enable both Service Providers and subscribers to optimize bandwidth and economics. Service Providers have the ability to offer bandwidth in small increments and usually without having to add new physical interfaces. This means they can offer, engineer, and bill only the bandwidth needed by the subscriber for a specific service. NOTE Multiple services can be offered over a subscriber, and each of these services can have its own bandwidth profile. Subscribers can purchase and pay for only the bandwidth they need. Furthermore, subscribers can be assured of a committed amount of bandwidth that meets certain performance objectives (usually specified in an SLA) and excess bandwidth that may not meet the SLA. 14 Half duplex means transmission in one direction at any one time. Full duplex means transmission in both directions simultaneously; these are briefly discussed in Chapter 1. ch02.indd 62 10/1/07 10:31:56 PM

19 Carrier Ethernet 63 Bandwidth Profile Traffic Parameters A Bandwidth profile associated with an Ethernet service consists of four traffic parameters: Committed Information Rate (CIR), Committed Burst Size (CBS), Excess Information Rate (EIR), and Excess Burst Size (EBS); in addition a service frame is associated with a Color Mode (CM). Together, these five parameters specify the bandwidth profile for a particular service: Bandwidth Profile = <CIR, CBS, EIR, EBS, CM> Committed Information Rate (CIR) CIR is the average rate up to which service frames are delivered as per the performance objectives (such as delay, loss, etc.) associated with the service; these service frames are referred to as being CIR-conformant. The CIR value is always less than or equal to the speed 15 and basically guarantees that the specified amount of bandwidth (or service frames) will be delivered according to a predetermined performance level. A CIR of zero indicates the service has neither bandwidth nor performance guarantees. NOTE Independent of the CIR, the service frames are always sent at speed. Committed Burst Size (CBS) CBS is the limit on the maximum number, or bursts, of service frames in bytes allowed for incoming service frames so they are still CIRconformant. Excess Information Rate (EIR) The EIR specifies the average rate, greater or equal to the CIR, up to which service frames are admitted into the Service Provider network; these frames are said to be EIR-conformant. These frames are delivered without any performance guarantees and are not CIR-conformant; however, service frames that are not EIR-conformant are discarded. Again, independent of the EIR, the service frames are always sent at the speed of the (and hence, the EIR represents the average rate). Excess Burst Size (EBS) The EBS is the limit on the maximum number, or bursts, of service frames in bytes allowed for incoming service frames so they are still EIRconformant Color Mode and Color Marking In addition to the bandwidth profile traffic parameters, there is also the concept of marking the service frames with a color. The color of a service frame is used to determine whether or not a particular service frame is in conformance with its bandwidth profile. A service marked green is conformant with the CIR and CBS in the bandwidth profile. A green frame is always delivered per the performance SLA associated with the service. 15 If multiple services are being delivered over a, then the sum of the CIRs associated with individual services must be less than or equal to the speed. ch02.indd 63 10/1/07 10:31:57 PM

20 64 Chapter 2 Yellow frames are out-of-bandwidth profile and will be delivered only if there are adequate bandwidth resources; if, on the other hand, the network is congested, then the frame is discarded. A red service frame is also out-of-bandwidth profile and is immediately discarded. The Color Mode (CM) parameter specifies whether the is operating in a coloraware or color-blind mode. When in a color-aware mode, the color associated with an incoming service frame is employed; in the color-blind mode, the color indication is ignored. Bandwidth Profile Rate Enforcement The Bandwidth profile is enforced through a two rate (committed or excess), three-color marker (green, yellow, or red) algorithm, referred to as the trtcm algorithm; this algorithm is usually implemented using a token bucket concept and is shown in Figure 2.7. Two buckets, one referred to as the committed or C-bucket and the other referred to as the excess or E-bucket, are used. Initially, each of these buckets is full of tokens; the C-bucket has green tokens and the E-bucket has yellow tokens. As service frames enter the Service Provider network, the same number of tokens in the C-bucket are removed (decreased). If, after this, there are green tokens in the C-bucket, then the service frame is CIR-conformant, colored green, and allowed in the network. If no green tokens remain, however, then the E-bucket is checked to determine if any yellow tokens remain. If there are yellow tokens, then the service frame is EIR-conformant, colored yellow, and allowed in the network. If no yellow tokens are available, then the service frame is colored red and discarded. Bandwidth Profile Defined by Token Bucket Algorithm (2 rates, 3 colors) Committed Information Rate (CIR) Excess Information Rate (EIR) Green Tokens Yellow Tokens Overflow Overflow Committed Burst Size (CBS) Excess Burst Size (EBS) C-Bucket E-Bucket Figure 2.7 Enforcing a predefined bandwidth profile using the token bucket concept (Source: MEF) ch02.indd 64 10/1/07 10:31:58 PM